Kong makes Kuma more K8s friendly, pushes observability forward


Service mesh and microservice control plane Kuma has seen the addition of a couple of observability features, bumping the project to version 0.4.

The current release of the Envoy-based project API and microservice tool provider Kong introduced in September 2019 should help it work better with other tools and brings a new TrafficTrace policy, amongst other things. The new addition lets users setup tracing for layer 7 HTTP traffic and can, for example, help to identify bottlenecks by monitoring the data exchange between services. 

Currently cloud native tools Zipkin and Jaeger are supported for such endeavours. Traffic traces can be found in the Kuma GUI, which has also been improved to load data quicker.

On another observability note, there are now three official dashboards for visualisation project Grafana available. Those are meant to help users gain insight into the traffic metrics collected by monitoring tool Prometheus, a staple in the cloud native realm.

Since having IP addresses for Gateway dataplanes wasn’t an option in earlier versions, Kuma 0.4 comes with a new dataplane format to both remediate this issue and simplify the interface format. It includes validation and is supposedly backwards compatible with its predecessor, so switching has its perks but isn’t absolutely necessary for now.

To make the control plane play nicer with Kubernetes, Kuma now emits events to let users know if a dataplane can’t be generated automatically for a pod. It also adds Kubernetes’ namespaces into label sets describing dataplanes to Prometheus and gives devs a chance to opt out of sidecar injection on Kubernetes by setting kuma.io/sidecar-injection: disabled.

Jumping onto the new version should be straightforward according to the upgrade document. More information on the changes can be found in the Kuma GitHub repository.